Charge-carrier transport simulations in diamond detectors with electric-field-dependent mobility and charge-collection-distance-based trapping
Paper details
Section titled “Paper details”| Field | Value |
|---|---|
| Journal | Nuclear Instruments and Methods in Physics Research Section A Accelerators Spectrometers Detectors and Associated Equipment |
| Published | 2026-10-19 |
| Authors | Faiz Ur Rahman Ishaqzai, Muhammed Deniz, Marta Baselga, Tobias Bisanz, Kevin Kröninger |
| DOI | 10.1016/j.nima.2026.172164 |
| Citations | 1 |
Allpix Squared reproduced single-crystal CVD drift velocities against a benchmark of 17 TCT experiments. The benchmark covered more than 45 samples and electric fields from 0.009 to 12 V/μm. The framework also reproduced the main features of polycrystalline CVD transients at 1 V/μm when measured CCD was supplied. The result links laboratory transport data to charge collection, pulse shape, and timing simulations.
Detector-level transport implementation
Section titled “Detector-level transport implementation”The implementation adds diamond to Allpix Squared’s Geant4-to-detector chain. Geant4 deposits energy, after which drift-diffusion propagation updates carrier position, stochastic diffusion, trapping, and induced charge. Mobility is evaluated from the local electric field during each integration step. The study used a spatially uniform field, although the framework can also evaluate nonuniform fields.
Electrons use the Piecewise model with a 0.08 V/μm transition field. Holes use the Caughey-Thomas model. The adopted transport parameters are:
| Carrier | Model | Low-field mobility (cm²/Vs) | Saturation velocity (10⁶ cm/s) | Characteristic field (V/μm) |
|---|---|---|---|---|
| Electron | Piecewise | 1880 ± 4 | 26.59 ± 0.51 | 0.56 |
| Hole | Caughey-Thomas | 2744 ± 9.4 | 15.09 ± 0.05 | 0.55 |
The shaping parameter β was 0.41 for electrons and 0.88 for holes. At 1 V/μm, simulated drift velocities were 64.4 μm/ns for electrons and 89.0 μm/ns for holes. These values agree with cited TCT measurements and the authors’ mobility review. All mobility parameters were fixed at central values, so their stated uncertainties did not propagate into the observables.
Charge collection distance as trapping input
Section titled “Charge collection distance as trapping input”Charge trapping is supplied through the charge collection distance, or CCD. Charge collection efficiency is CCE = Q/Q₀, and CCD = d × CCE. The Hecht relation converts CCD into effective electron and hole mean free paths for ionization averaged over sensor depth.
For unirradiated single-crystal CVD diamond, CCD is usually about d because trapping is negligible. For polycrystalline CVD diamond with CCD below d, CCD is about λₑ + λₕ under the model’s averaging assumptions.
The code uses r = λₕ/λₑ, with a default of 1.3 adopted from prior work. It solves for λₑ and then sets λₕ = rλₑ. Allpix Squared applies an exponential trapping probability over each distance step.
Signal loss from defects, grain boundaries, or irradiation is evaluated outside the transport code and represented by CCD. The model captures integral charge loss without identifying individual trap types.
Reference detector and measurements
Section titled “Reference detector and measurements”The pcCVD reference was an Element Six planar detector with 500 μm thickness. Its sensor area was 10 × 10 mm² and its electrodes measured 8 × 8 mm². Pumping used a ⁹⁰Sr beta source for approximately 20 minutes before bias application.
CCD measurements used ⁹⁰Sr beta particles and a downstream scintillator trigger. Each bias stored 2,000 waveforms, with SNR > 3 and at least 1,700 accepted events required. The Landau-Gaussian most-probable value supplied the collected charge, and an assumed 36 pairs/μm yield set the absolute CCD scale.
TCT measurements used a ²⁴¹Am alpha source incident from the grounded side. The CIVIDEC C2-TCT amplifier provided 40 dB gain and a 10 kHz to 2 GHz bandwidth. Waveforms were digitized at 10 GS/s.
IV measurements found leakage below approximately 0.4 nA from -1,000 V to +1,000 V. Capacitance remained approximately 6.127 pF, 5.2% below the 6.46 pF parallel-plate estimate. This stable terminal response did not establish spatial uniformity of the internal electric field.
Single-crystal mobility validation
Section titled “Single-crystal mobility validation”The scCVD benchmark combined 17 TCT experiments covering more than 45 samples over 0.009 to 12 V/μm. It included alpha and laser excitation. The authors pooled the data because sample quality, surface termination, contacts, excitation, and field window produce a large parameter spread.
Validation used time of arrival, transient FWHM, and direct waveform comparisons. Time-of-arrival simulations covered 60 bias values from -1 V to -40,000 V. The transient scan contained 34 electron-field settings and 29 hole-field settings.
At 0.18 V/μm, the simulated waveform was compared with a 526 μm scCVD sample. The simulation rose within the first nanosecond, held an approximately constant plateau, and fell rapidly. Measured edges were broader because the experimental transfer function was not fully included.
The unavailable full detector-amplifier-cable-oscilloscope transfer function prevents assigning late negative excursions conclusively.
Polycrystalline trapping validation
Section titled “Polycrystalline trapping validation”The measured CCD ranged from approximately 205 to 235 μm at 0.2 V/μm and entered quasi-saturation above 0.6 V/μm. Negative-bias CCD was typically 15 to 30 μm higher than positive-bias CCD. Pumping improved collection in both polarities.
The authors suggest stronger hole trapping near the nucleation side as an explanation. They did not establish that mechanism directly.
At 1 V/μm, measured CCD entered the trapping simulation. The simulation reproduced the principal amplitude scale, time of maximum, and overall duration for electrons and holes.
The hole waveform had a sharp falling-edge cusp. The electron waveform had lower amplitude and a broader decay. With λₕ/λₑ = 1.3, electrons received a shorter effective trapping length and greater charge loss over comparable paths.
The simulated electron leading edge remained steeper than measured, although the trailing edge followed most of the fall. The hole rise was also too abrupt, and its maximum occurred slightly earlier and higher than measured. Residual late-time differences remained after the main pulse.
Model limits and selection implications
Section titled “Model limits and selection implications”The model uses an effective bulk description rather than resolved defect or grain physics. It omits polarization, nonuniform internal fields, spatially varying trapping, detrapping, electrode-edge effects, and explicit contacts.
The dense alpha-generated cloud may transiently screen the applied field. The authors could not separate that effect from bandwidth, diffusion, trapping, and field inhomogeneity. A complete electronics impulse response was absent, so bipolar undershoot and overshoot were not reproduced.
The assumed 36 pairs/μm yield has an unquantified systematic effect on CCD. Gain calibration, thickness tolerance, and residual baseline systematics were also excluded from reported CCD errors. Mobility uncertainties were fixed rather than propagated.
For scCVD, material selection can start with CCD ≈ d and the validated mobility models when trapping is negligible. For pcCVD, CCD should be measured versus field and polarity because grain boundaries and carrier-dependent trapping affect collection. Pumped and unpumped states also need separate CCD inputs.
The framework is suitable for signal, timing, and charge collection studies when its inputs come from material measurements. Radiation-damage predictions still require validation over more thicknesses, bias conditions, and irradiation-dependent CCD parameterizations. The current trapping demonstration uses a 500 μm pcCVD detector at 1 V/μm.
Materials and support
Section titled “Materials and support”For planar particle detectors, the validation hardware used pcCVD diamond with metal electrodes and a defined thickness. The diamond materials catalog covers MPCVD-grown single-crystal and polycrystalline plates and wafers up to 125 mm, with thicknesses from 0.1 to 500 micrometers and substrates up to 10 mm. Detector fabrication questions can be directed to 6CCVD’s engineering support, with Au, Pt, Pd, Ti, W, or Cu metallization available.
View Original Abstract
Diamond detectors are attractive for operation in harsh radiation environments because they combine radiation tolerance, fast signal formation, and low leakage current. Realistic detector-response simulations require an accurate description of charge-carrier mobility and trapping, which determine both signal amplitude and timing. In this work, we extend Allpix Squared, a modular end-to-end detector simulation framework, with diamond-specific transport models. The implementation includes field-dependent mobility parameterizations for electrons and holes and an effective trapping model based on the charge collection distance (CCD), providing a detector-level interface to material quality and radiation-damage measurements. The mobility description is validated in the negligible-trapping limit using single-crystalline CVD diamond by comparing simulated drift velocities and transient-current signals with published reference data. For polycrystalline CVD diamond, the CCD-based trapping model is evaluated using experimentally measured CCD values and compared with laboratory transient-current-technique waveforms. The simulations reproduce the drift-velocity behavior of the scCVD reference data and, with the measured CCD as input, the main features of the degraded transient response observed in pcCVD. The presented implementation enables detector-level studies of charge collection, pulse formation, and timing performance in diamond sensors using experimentally accessible transport and trapping parameters, and provides a practical framework for simulation-driven detector development and radiation-damage studies.
Single Crystal Cvd Diamond Plate
View Specs →
Customized Single Crystal Cvd Diamond Plate
View Specs →